Automation/app/Libraries/MyCrawler/Mangboard/InvenCrawler.php

<?php

namespace App\Libraries\MyCrawler\Mangboard;

use App\Entities\Mangboard\UserEntity;
use Symfony\Component\DomCrawler\Crawler;

class InvenCrawler extends MangboardCrawler
{
    public function __construct(string $host, string $board_name, UserEntity $user_entity)
    {
        parent::__construct($host, $board_name, $user_entity);
    }
    protected function getUrlByMediaType(Crawler $node, string $media_type, string $attr): null|string
    {
        switch ($media_type) {
            case 'video':
                $url = parent::getUrlByMediaType($node, $media_type, $attr);
                //그래도 null이면 data-src로 추출해본다.
                $attributes = $node->extract(['data-src']);
                if (count($attributes)) {
                    $url = $attributes[0];
                }
                break;
            case 'img':
            default:
                $url = parent::getUrlByMediaType($node, $media_type,  $attr);
                break;
        }
        return $url;
    }
    //작성내용
    // <div class="articleContent">
    // <div id="imageCollectDiv" class="contentBody">
    //     <!-- ============== CONTENT ============== -->
    //     <div id="powerbbsContent">
    //         <div id="BBSImageHolderTop" style="text-align:center;">
    //              <img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1620925350.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 1080 / 1350;" loading="lazy" />
    //              <br><br><img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1587803007.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 1080 / 1350;" loading="lazy" />
    //              <br><br><img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1134295360.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 1080 / 1350;" loading="lazy" />
    //              <br><br><img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1481352611.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 1080 / 1350;" loading="lazy" />
    //              <br><br><img src="https://upload3.inven.co.kr/upload/2024/09/15/bbs/i1878651605.jpg?MW=800" style="max-width: 100%; width: 800px; aspect-ratio: 850 / 1063;" loading="lazy" />
    //              <br><br>
    //         </div>
    //         <div>^^</div>
    //     </div>
    //     <!-- ============== End CONTENT ============== -->
    // </div>    
    protected function detail_page(int $cnt, array $listInfo): array
    {
        $response = $this->getMySocket()->getContent($listInfo['detail_url']);
        $tag = getenv("inven.view.content.tag");
        $selector = $this->getSelector($response, $tag);
        $media_urls = $this->getUrlsByMediaType($selector, "img",  "src");
        $media_urls = $this->getUrlsByMediaType($selector,  "video",  "src", $media_urls);
        if ($this->getDebug()) {
            throw new \Exception(sprintf(
                "\n--------------%s Debug--------------\n%s%s\n---------------------------------------\n",
                __FUNCTION__,
                var_export($listInfo, true),
                var_export($media_urls, true)
            ));
        } else {
            // Image 나 Video 소스들의 url을 가져와서 실제 다운받는 처리
            $storages = $this->media_process($media_urls);
            if (!count($storages)) {
                throw new \Exception("등록할 자료가 없습니다.");
            }
            $this->backend_process($cnt, $listInfo, $storages);
        }
        return $listInfo;
    }
    //리스트내용
    // <div class="board-list">
    //     <table>
    //         <tr class="lgtm">
    //             <td class="num"><span>1589</span></td>
    //             <td class="tit">
    //                 <div class="text-wrap">
    //                     <div>
    //                         <span class="user-icon">
    //                             <img src="https://upload3.inven.co.kr/upload/2024/06/12/icon/i1237935053.jpg" alt="유저 아이콘" loading="lazy">
    //                         </span>
    //                         <a class="subject-link" href="https://www.inven.co.kr/board/party/5951/1589">
    //                             <span class="board_name">[사진＆움짤]</span>스테이씨 윤
    //                         </a>
    //                     </div>
    //                     <span data-opinion-bbs-comeidx="5951" data-opinion-bbs-uid="1589" data-opinion-bbs-opi="1" class="con-comment">[1]</span>                                                        
    //                     <span class="con-icon board-img photo">사진</span>
    //                 </div>
    //             </td>
    //             <td class="user">
    //                 <img src="https://static.inven.co.kr/image_2011/member/level/1202/lv32.gif" alt="레벨 아이콘">
    //                 <span class="layerNickName" onclick="layerNickName('배수민', 'pbNickNameHandler'); ">배수민</span>                                            
    //             </td>
    //             <td class="date">09-15</td>
    //             <td class="view">1,502</td>
    //             <td class="reco">1</td>
    //         </tr>
    //     </table>
    // </div>
    public function execute(int $max_limit): void
    {
        try {
            if ($this->getDebug()) {
                $url = getenv("inven.view.test.url.{$this->_board_name}");
                log_message("notice", __FUNCTION__ . "=> DEBUG 게시물 {$url} 작업시작");
                $this->detail_page(1, ['detail_url' => $url]);
                log_message("notice", __FUNCTION__ . "=> DEBUG 게시물 {$url} 작업종료");
            } else {
                $listInfos   = [];
                $response = $this->getMySocket()->getContent(getenv("inven.list.url.{$this->_board_name}"));
                $this->getSelector($response, getenv("inven.list.tag"))->each(
                    function (Crawler $node) use (&$listInfos): void {
                        $hit      = $node->filter(getenv("inven.list.item.hit.tag"))->text();
                        $date     = date("Y") . "-" . $node->filter(getenv("inven.list.item.date.tag"))->text();
                        $nickname = $node->filter(getenv("inven.list.item.nickname.tag"))->text();
                        //작성자가 "관리자"가 아닌 게시물이면 해당 bbs_item에서 a.list_subject 객체를 찾아서
                        $link_node   = $node->filter(getenv("inven.list.item.link.tag"));
                        $detail_url  = $link_node->attr("href");
                        $title       = $link_node->text();
                        $listInfos[] = ['title' => $title, 'nickname' => $nickname, 'detail_url' => $detail_url, 'date' => $date, 'hit' => $hit];
                    }
                );
                if (!count($listInfos)) {
                    throw new \Exception("Target URL이 없습니다.");
                }
                $this->main_process($max_limit, $listInfos);
            }
            log_message("notice",   __FUNCTION__ . " 작업이 완료되었습니다.");
        } catch (\Exception $e) {
            log_message("warning", sprintf(
                "\n---%s 오류---\n%s\n-----------------------------------------\n",
                __FUNCTION__,
                $e->getMessage()
            ));
        }
    }
}